Silicon Valley billionaires bankrolling the Artificial Intelligence boom are also manipulating its perception — sponsoring research and testing labs, ambassadors and the journalists reporting on it.
The most impactful operation is by those involved with the Effective Altruism (EA) movement, who have created an entire ecosystem which has propelled Anthropic from startup in 2021 to the forefront of AI and a $965 billion valuation today.
Backers such as Dustin Moskovitz and now-imprisoned Sam Bankman-Fried poured money not just into the AI company, but also labs testing AI safety and reporting on the industry.
Critics say the overlapping money has built an echo chamber, amplifying Anthropic and its CEO Dario Amodei’s doomerist message. But some say that message is likely tactical.
“I think Dario seems to be using anxiety about safety as a weapon to control the pace of business,” Nathan Zhao, founding partner of tech investment firm SevenUp Capital, told The Post.
“In Anthropic’s early days, there was a strong cult-like culture. It started with a shared belief in putting safety first, but that could easily turn into employees having unquestioning faith in everything Dario said,” claimed Zhao, who’s spoken with several Anthropic employees.
Industry insiders claimed Amodei’s calls for government regulation of AI this week are self-serving, as smaller companies would likely not have the funds to comply, upsetting many.
“What is Anthropic actually trying to achieve? Nobody knows. But this attitude is really irritating — they’re always claiming the moral high ground and criticizing everyone else,” said Richard Chang, an AI engineer who asked to withhold his employer’s details.
“[Until recently] people generally had a very positive impression of Anthropic. But people in the industry have started reacting very negatively to the various positions they’ve taken.”
The underlying message is Anthropic is the only company to be trusted and their way forward is the safest, a message being amplified through its network, which can be traced back to the company’s founding.
Moskovitz, formerly of Facebook, took part in Anthropic’s $124 million funding round in 2021, which was led by Skype co-founder Jaan Tallinn.
Moskovitz also sits on the board of Coefficient Giving, a nonprofit previously known as Open Philanthropy, which he set up with wife Cari Tuna.
Coefficient is also a major backer of the Tarbell Center for AI Journalism, which pays the salaries of over 80 reporters embedded to cover the industry at outlets including Time, Bloomberg, The Guardian and The Verge. The center stresses its journalists are independent.
Meanwhile, another EA group, Longview Philanthropy — a nonprofit with $10 million in revenue according to government filings — funds Model Evaluation and Threat Research (METR), a company which tests AI models’ benchmarks as well as existential threats. That company’s board has close ties to Anthropic and rivals OpenAI through its head, Paul Christiano, who used to be Amodei’s roommate. They both previously worked at OpenAI.
Finally, Anthropic pays evangelists in the field, recruiting and paying the salaries of 1,000 people to be the trained on its AI assistant Claude and embed in over 400 nonprofit organizations as the “Claude Corps.” They also pay Claude College Ambassadors and recruit Claude Community Ambassadors.
“Before, they were very public that the goal of many of these programs was to promote Effective Altruism,” Brian Chau, a machine learning engineer and founder of AI news site Effort, told the Post.
That all changed with the December 2022 arrest of Sam Bankman-Fried, the now-disgraced founder of crypto exchange FTX and one of EA’s most prominent champions.
Chau said people in effective-altruism circles described Bankman-Fried as a mentor to Amodei before FTX collapsed, and regarded the association as a badge of honor.
Chau argues that Amodei is now following Bankman-Fried’s approach to regulation: calling for safety rules that could also favor an established company.
“So it’s no surprise that he’s now copying literally the exact same tactics used by Sam Bankman-Fried in order to achieve regulatory monopoly,” Chau told The Post.
Bankman-Fried led Anthropic’s $580 million funding round in April 2022, months before his crypto empire collapsed and he was prosecuted for scamming investors to the tune of $3 billion. Convicted of fraud and conspiracy, he is now serving 25 years in prison.
“The description of the program 1789640155 is to ‘promote AI safety.’ They’ve stayed away from the term Effective Altruism ever since Sam Bankman-Fried got arrested,” Chau said.
Federal prosecutors forced the sale of Bankman-Fried’s stake in Anthropic and other companies to repay some of his debts. He is no longer associated or has a stake in any technology company.
Chau said those involved in EA are working their way into a difficult position.
“The more the incentives make it so that you benefit from some kind of belief, the less you will know that’s why you’re believing it,” he said. “I think that’s true of Dario, and I think that’s true of most of these Effective Altruists.
“They don’t fully realize the reason they believe the world is going to end is because [of the same thing] that gives them status in their social circles and, for many of them, advantages economically.”
His point is echoed by a longtime follower of the EA movement, which has its roots in charity giving and philosophers William MacAskill and Peter Singer, but was only formalized over the last 15 years.
“I don’t like the switch in focus from global poverty and global health, fundraising for charities like the Against Malaria Foundation, to now a focus on existential fears about AI. That’s the main problem,” the source, who asked to remain anonymous, told The Post.
